Loading history
Last graded 2026-09-12
Every assessment is published before the event and graded automatically from the official result. Once a card starts, nothing about it can change. The system learns from its graded history.
Record
17–10
27 graded
Winner accuracy
63%
The named winner won
Cards graded
2
Completed cards with a graded bout
Fights graded
27
Decided assessments
Current streak
3 correct in a row
Consecutive graded assessments
Main event accuracy
100%
2 of 2 main events
Probability calibration
0.239
Brier score · lower is better · 0.25 = always saying 50%
Once a card is sealed the sheet and the assessments are frozen. Completed cards are never reopened.
Accuracy asks whether the named winner won. Confidence asks whether the percentages were honest: bouts assessed at around 60% should be correct around 60% of the time.
Each dot is a group of assessments with similar percentages. On the dashed line, the percentages matched the results. Below it, the assessments were too confident; above it, too cautious.
Brier score: how well the stated probabilities matched the results. Lower is better; 0.25 is what saying 50% on every bout would score.
| Confidence | Said | Correct | Assessments |
|---|---|---|---|
| 50%–55% | 54% | 100% | 1 |
| 55%–60% | 57% | 53% | 19 |
| 60%–70% | 62% | 100% | 6 |
| 70%–80% | 72% | 0% | 1 |
| Card | Date | Record | Accuracy |
|---|---|---|---|
| Noche UFC: Silva vs. Delgado 2026-09-12 | 8–5 | 62% | |
| UFC Fight Night: Hooker vs. Parnasse 2026-09-05 | 9–5 | 64% |
The named winner and how they were expected to win, counted correct only when that fighter won that way.
No winner-and-finish assessments have been published yet. This section fills in once a card is assessed that way.
When a finish type was named, how often the fight ended that way, whoever won.
44%
0%
59%